U.S. patent application number 13/728572 was filed with the patent office on 2014-07-03 for methods and systems for identifying a precursor to a failure of a component in a physical system.
This patent application is currently assigned to GENERAL ELECTRIC COMPANY. The applicant listed for this patent is GENERAL ELECTRIC COMPANY. Invention is credited to James Kenneth Aragones, Piero Patrone Bonissone, Naresh Sundaram Iyer, Brock Estel Osborn, Hai Qiu, Anil Varma, Weizhong Yan.
Application Number | 20140188777 13/728572 |
Document ID | / |
Family ID | 51018352 |
Filed Date | 2014-07-03 |
United States Patent
Application |
20140188777 |
Kind Code |
A1 |
Yan; Weizhong ; et
al. |
July 3, 2014 |
METHODS AND SYSTEMS FOR IDENTIFYING A PRECURSOR TO A FAILURE OF A
COMPONENT IN A PHYSICAL SYSTEM
Abstract
A computer-implemented system for identifying a precursor to a
failure of a particular type of component in a physical system is
provided. The physical system includes sensors coupled to the
physical system. The computer-implemented system includes a
computing device, a database, a processor, and a memory device. The
memory device includes historical data including sensor
measurements. When instructions are executed by the processor, the
processor receives the historical data from the memory device. The
processor generates a predictive model. The predictive model uses,
as inputs, sensor measurements in the historical data. The
predictive model is able to differentiate between sensor
measurements taken before the repair event and those taken after
the repair event without a time of the repair event being an input
to the predictive model. The processor designates at least one
sensor measurements used as inputs to the predictive model as
precursors to the failure of the component.
Inventors: |
Yan; Weizhong; (Clifton
Park, NY) ; Varma; Anil; (San Ramon, CA) ;
Osborn; Brock Estel; (Niskayuna, NY) ; Aragones;
James Kenneth; (Clifton Park, NY) ; Bonissone; Piero
Patrone; (Schenectady, NY) ; Iyer; Naresh
Sundaram; (Saratoga Springs, NY) ; Qiu; Hai;
(Clifton Park, NY) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
GENERAL ELECTRIC COMPANY |
Schenectady |
NY |
US |
|
|
Assignee: |
GENERAL ELECTRIC COMPANY
Schenectady
NY
|
Family ID: |
51018352 |
Appl. No.: |
13/728572 |
Filed: |
December 27, 2012 |
Current U.S.
Class: |
706/50 |
Current CPC
Class: |
G06F 11/008 20130101;
G06N 5/04 20130101 |
Class at
Publication: |
706/50 |
International
Class: |
G06N 5/02 20060101
G06N005/02 |
Claims
1. A computer-implemented system for identifying a precursor to a
failure of a particular type of component in a physical system, the
physical system having a plurality of sensors coupled to components
of the physical system, said computer-implemented system
comprising: a computing device; a database associated with said
computing device; a processor coupled to said computing device; and
a memory device coupled to said processor and to said computing
device, said memory device including historical data including
sensor measurements from the plurality of sensors over a time
period, wherein the time period at least spans operation of a
replaced component of the particular type immediately preceding and
immediately following a repair event in which the replaced
component failed and was replaced, said memory device further
including processor-executable instructions that, when executed by
said processor, cause said processor to: receive the historical
data from said memory device; generate a predictive model which
uses, as inputs, sensor measurements in the historical data,
wherein said predictive model is able to differentiate between
sensor measurements taken before the repair event and sensor
measurements taken after the repair event, without a time of the
repair event being an input to the predictive model; and designate
at least one sensor measurement used as inputs to said predictive
model as precursors to the failure of the particular type of
component.
2. The computer-implemented system of claim 1, wherein said memory
device including processor-executable instructions to generate said
predictive model further including process-executable instructions
to generate said predictive model which uses, as inputs, a subset
of the sensor measurements in the historical data.
3. The computer-implemented system of claim 1, wherein said memory
device further includes processor-executable instructions to
generate said predictive model such that said predictive model that
combines at least two sensor measurements in a mathematical
operation.
4. The computer-implemented system of claim 1, wherein said memory
device further includes historical data including sensor
measurements from the plurality of sensors over a time period,
where the time period includes multiple repair events.
5. The computer-implemented system of claim 4, wherein said memory
device further includes processor-executable instructions that,
when executed, cause said processor to generate said predictive
model and to further: generate a plurality of candidate predictive
models, wherein said plurality of candidate predictive models use a
random selection of sensor measurements as inputs; determine which
of said plurality of candidate predictive models most accurately
differentiates between sensor measurements taken before the repair
events and sensor measurements taken after the repair events; and
designate, as said predictive model, a most accurate candidate
predictive model from said plurality of candidate predictive
models.
6. The computer-implemented system of claim 4, wherein said memory
device further includes processor-executable instructions that,
when executed by said processor, designates the at least one sensor
measurement as precursors to the failure of the particular type of
component cause said processor to further: designate a combination
of sensor measurements, combined with at least one mathematical
operation of the predictive model, as a precursor to failure; and
transmit the designated combination of sensor measurements and the
at least one mathematical operation of the predictive model to a
monitoring system, the monitoring system monitors the physical
system.
7. The computer-implemented system of claim 1, wherein said memory
device further includes processor-executable instructions that,
when executed, cause said processor to receive the historical data
from said memory device and to further: receive expert data from
said memory device, the expert data associated with the historical
data, the expert data substantially representing information
obtained by human experts regarding the relationship between the
plurality of sensors and the physical system; and generate said
predictive model using at least some of the expert data.
8. A computer-implemented method for identifying a precursor to a
failure of a particular type of component in a physical system, the
physical system having a plurality of sensors coupled to components
of the physical system, the method is performed by a computing
device including: a processor coupled to a memory device and
associated with a database, said memory device including historical
data including sensor measurements from the plurality of sensors
over a time period, wherein the time period at least spans an
operation of a replaced component of the particular type
immediately preceding and immediately following a repair event in
which the replaced component failed and was replaced, said method
comprising: receiving the historical data from said memory device;
generating a predictive model which uses, as inputs, sensor
measurements in the historical data, wherein the predictive model
is able to differentiate between sensor measurements taken before
the repair event and sensor measurements taken after the repair
event, without a time of the repair event being an input to the
predictive model; and designating at least one sensor measurement
used as inputs to the predictive model as precursors to the failure
of the particular type of component.
9. The computer-implemented method of claim 8, wherein generating
said predictive model comprises generating a predictive model which
uses, as inputs, a subset of the sensor measurements in the
historical data.
10. The computer-implemented method of claim 8, wherein generating
said predictive model comprises generating a predictive model that
combines at least two sensor measurements in a mathematical
operation.
11. The computer-implemented method of claim 8, wherein the time
period includes multiple repair events.
12. The computer-implemented method of claim 11, wherein generating
the predictive model comprises: generating a plurality of candidate
predictive models, wherein said plurality of candidate predictive
models use a random selection of sensor measurements as inputs;
determining which of said plurality of candidate predictive models
most accurately differentiates between sensor measurements taken
before the repair events and sensor measurements taken after the
repair events; and designating, as said predictive model, a most
accurate candidate predictive model from said plurality of
candidate predictive models.
13. The computer-implemented method of claim 11, wherein
designating the at least one sensor measurement as precursors to
the failure of the particular type of component further comprises:
designating a combination of sensor measurements, combined with at
least one mathematical operation of the predictive model, as a
precursor to the failure; and transmitting the designated
combination of sensor measurements and the at least one
mathematical operation of the predictive model to a monitoring
system, the monitoring system monitors the physical system.
14. The computer-implemented method of claim 8, further comprising:
receiving expert data from said memory device, the expert data
associated with the historical data, the expert data substantially
representing information obtained by human experts regarding the
relationship between the plurality of sensors and the physical
system; and generating said predictive model using at least some
the expert data.
15. A computer-readable storage device having processor-executable
instructions embodied thereon, wherein, when executed by at least
one processor coupled to a memory device in a computing device,
said memory device including historical data including sensor
measurements from a plurality of sensors over a time period, the
time period at least spanning operation of a replaced component of
a particular type immediately preceding and immediately following a
repair event in which the replaced component failed and was
replaced, cause the at least one processor to: receive the
historical data from said memory device; generate a predictive
model which uses, as inputs, sensor measurements in the historical
data, wherein the predictive model is able to differentiate between
sensor measurements taken before the repair event and sensor
measurements taken after the repair event, without a time of the
repair event being an input to the predictive model; and designate
at least one sensor measurement used as inputs to the predictive
model as precursors to the failure of the particular type of
component.
16. The computer-readable storage device of claim 15, wherein said
computer-readable storage device further has processor-executable
instructions that generate a predictive model such that the
predictive model uses, as inputs, a subset of the sensor
measurements in the historical data.
17. The computer-readable storage device of claim 15, wherein said
computer-readable storage device further has processor-executable
instructions that generate a predictive model such that the
predictive model combines at least two sensor measurements in a
mathematical operation.
18. The computer-readable storage device of claim 15, wherein the
time period includes multiple repair events.
19. The computer-readable storage device of claim 18, wherein said
computer-readable storage device further has processor-executable
instructions that generate further has processor-executable
instructions that: generate a plurality of candidate predictive
models, wherein each predictive model uses a random selection of
sensor measurements as inputs; determine which of the plurality of
candidate predictive models most accurately differentiates between
sensor measurements taken before the repair events and sensor
measurements taken after the repair events; and designate the most
accurate candidate predictive model as the predictive model.
20. The computer-readable storage device of claim 18, wherein said
computer-readable storage device further has processor-executable
instructions that generate further has processor-executable
instructions that: designate a combination of sensor measurements,
combined with at least one mathematical operation of the predictive
model, as a precursor to failure; and transmit the designated
combination of sensor measurements and the at least one
mathematical operation of the predictive model to a monitoring
system, the monitoring system monitors the physical system.
Description
BACKGROUND
[0001] The field of the invention relates generally to maintenance
of components of a physical system, and more particularly, to a
computer-implemented system for identifying a precursor to a
failure of a particular type of component in a physical system.
[0002] Many known complex physical systems, such as aircraft,
automobiles, and physical systems used in industrial plants,
include multiple components that perform repetitive functions. Over
time, it is possible for the components to wear such that they
approach the end of useful life. In many instances, sensors are
included within, coupled to, or otherwise in the vicinity of a
physical system and electronically transmit sensor measurements,
i.e., measurement data determined by the sensor, to a central
computing device for evaluation. For many components, the set of
sensors or measurements that carry information related to the
component's health and thus remaining useful life might be
previously known. For example, increasing vibration sensor
measurements collected by a sensor over a given time period may be
used to infer that a particular bearing in a physical system is
wearing out and will approach the end of useful life within a
month. However, for many other components, the existing
measurements or sensors that carry information related to the
health or degradation of the component might not be known a priori.
In fact, one needs to look at the entire potential set of sensor
measurements and construct or synthesize the health of the
component using advanced models that map these diverse set of
sensors to component health. This process of constructing such a
model is extremely complex due to many factors including the amount
of data involved, the need to select the relevant subset of sensors
to use in the modeling from a long and combinatorially complex list
and the complexity of modeling approaches that have to be used.
[0003] In other known complex physical systems, sensors included
within, coupled to, or otherwise in the vicinity of the physical
system electronically send sensor measurements to a central
computing device for programmatic evaluation. While many known
software programs implementing a programmatic evaluation have the
ability to process significant amounts of data, many software
programs lack the knowledge of human experts regarding the sensors
measurements and the interactions between sensor measurements to be
used for estimating the useful life of the component. As a result,
these other known complex physical systems may process sensor
measurements but lack an ability to detect the sensor measurements
most associated with the failure of a component. Although some
known software programs can be taught to look for specific sensor
measurements, such some known software programs are dependent upon
a domain of knowledge, i.e., the area of expert knowledge specific
to a field of inquiry that is utilized for particular sensor
measurement analysis.
BRIEF DESCRIPTION
[0004] In one aspect, a computer-implemented system for identifying
a precursor to a failure of a particular type of component in a
physical system is provided. The physical system includes a
plurality of sensors coupled to components of the physical system.
The computer-implemented system includes a computing device, a
database associated with the computing device, a processor coupled
to the computing device, and a memory device coupled to the
processor and the computing device. The memory device includes
historical data including sensor measurements from the plurality of
sensors over a time period. The time period at least spans the
operation of a replaced component of the particular type
immediately preceding and immediately following a repair event in
which the replaced component failed and was replaced. The memory
device further includes processor-executable instructions. When the
processor-executable instructions are executed by the processor,
the processor receives the historical data from the memory device.
The processor then generates a predictive model. The predictive
model uses, as inputs, sensor measurements in the historical data.
The predictive model is able to differentiate between sensor
measurements taken before the repair event and sensor measurements
taken after the repair event without a time of the repair event
being an input to the predictive model. The processor then
designates at least one sensor measurements used as inputs to the
predictive model as precursors to the failure of the particular
type of component.
[0005] In another aspect, a computer-implemented method for
identifying a precursor to a failure of a particular type of
component in a physical system is provided. The physical system
includes a plurality of sensors coupled to components of the
physical system. The method is performed by a computing device. The
computing device includes a processor coupled to a memory device
and is associated with a database. The memory device includes
historical data including sensor measurements from the plurality of
sensors over a time period. The time period at least spans an
operation of a replaced component of the particular type
immediately preceding and immediately following a repair event in
which the replaced component failed and was replaced. The method
includes receiving the historical data from the memory device. The
method further includes generating a predictive model which uses as
inputs sensor measurements in the historical data. The predictive
model is able to differentiate between sensor measurements taken
before the repair event and sensor measurements taken after the
repair event without a time of the repair event being an input to
the predictive model. The method additionally includes designating
at least one sensor measurement used as inputs to the predictive
model as precursors to the failure of the particular type of
component.
[0006] In another aspect, a computer-readable storage device having
processor-executable instructions embedded thereon is provided. At
least one processor coupled to a memory device in a computing
device may execute the processor-executable instructions embedded
on the computer-readable storage device. The memory device includes
historical data including sensor measurements. The sensor
measurements are received from a plurality of sensors over a time
period. The time period at least spans the operation of a replaced
component of a particular type immediately preceding and
immediately following a repair event in which the replaced
component failed and was replaced. The processor receives the
historical data from the memory device. When the
processor-executable instructions are executed, the processor
generates a predictive model which uses, as inputs, sensor
measurements in the historical data. The predictive model is able
to differentiate between sensor measurements taken before the
repair event and sensor measurements taken after the repair event
without a time of the repair event being an input to the predictive
model. Also, when executed, the processor designates one or more
sensor measurements used as inputs to the predictive model as
precursors to the failure of the particular type of component.
DRAWINGS
[0007] These and other features, aspects, and advantages will
become better understood when the following detailed description is
read with reference to the accompanying drawings in which like
characters represent like parts throughout the drawings,
wherein:
[0008] FIG. 1 is a simplified block diagram of a portion of an
exemplary computer-based system for identifying a precursor to a
failure of a particular type of component in a physical system;
[0009] FIG. 2 is a block diagram of an exemplary computing device
that may be used in the computer-based system shown in FIG. 1;
[0010] FIG. 3 is a flow chart of an exemplary process of the flow
of information in the computer-based system shown in FIG. 1;
and
[0011] FIG. 4 is a flow chart of an exemplary method for
identifying a precursor to a failure of a particular type of
component in a physical system used in the computer-based system,
shown in FIG. 1, using the process shown in FIG. 3.
[0012] Unless otherwise indicated, the drawings provided herein are
meant to illustrate key inventive features of the invention. These
key inventive features are believed to be applicable in a wide
variety of systems comprising one or more embodiments of the
invention. As such, the drawings are not meant to include all
conventional features known by those of ordinary skill in the art
to be required for the practice of the invention.
DETAILED DESCRIPTION
[0013] In the following specification and the claims, reference
will be made to a number of terms, which shall be defined to have
the following meanings.
[0014] The singular forms "a", "an", and "the" include plural
references unless the context clearly dictates otherwise.
[0015] "Optional" or "optionally" means that the subsequently
described event or circumstance may or may not occur, and that the
description includes instances where the event occurs and instances
where it does not.
[0016] As used herein, the terms "software" and "firmware" are
interchangeable, and include any computer program stored in memory
for execution by devices that include, without limitation, mobile
devices, clusters, personal computers, workstations, clients, and
servers.
[0017] As used herein, the term "non-transitory computer-readable
media" is intended to be representative of any tangible
computer-based device implemented in any method or technology for
short-term and long-term storage of information, such as,
computer-readable instructions, data structures, program modules
and sub-modules, or other data in any device. Therefore, the
methods described herein may be encoded as executable instructions
embodied in a tangible, non-transitory, computer readable medium,
including, without limitation, a storage device and/or a memory
device. Such instructions, when executed by a processor, cause the
processor to perform at least a portion of the methods described
herein. Moreover, as used herein, the term "non-transitory
computer-readable media" includes all tangible, computer-readable
media, including, without limitation, non-transitory computer
storage devices, including, without limitation, volatile and
nonvolatile media, and removable and non-removable media such as a
firmware, physical and virtual storage, CD-ROMs, DVDs, and any
other digital source such as a network or the Internet, as well as
yet to be developed digital means, with the sole exception being a
transitory, propagating signal.
[0018] As used herein, the term "computer" and related terms, e.g.,
"computing device," are not limited to integrated circuits referred
to in the art as a computer, but broadly refers to a
microcontroller, a microcomputer, a programmable logic controller
(PLC), an application specific integrated circuit, and other
programmable circuits, and these terms are used interchangeably
herein.
[0019] As used herein, the term "physical system" and related
terms, e.g., "physical systems," refers to any system composed of
one or more parts that has a physical presence. Physical systems
may include, without limitation, vehicles, transportation systems,
manufacturing facilities, chemical processing facilities, power
generation facilities, infrastructure systems, and communication
systems. Physical systems may also include, without limitation,
complex chemical or biological systems where components of such
systems may have sensor measurements associated. Also, as used
herein, physical systems are analyzed to find precursors to failure
of a particular type of component of the physical system.
[0020] As used herein, the term "failure" and related terms, e.g.,
"failure incidents," means falling below the desired level of
performance. Failure does not require a physical breakdown or
adverse consequences for the physical system. Also, as used herein,
failure may refer to a particular type of component or a plurality
of components not meeting the expected level of performance.
[0021] As used herein, the term "precursor" and related terms,
e.g., "failure precursor," means a condition that is known or
expected to indicate a subsequent outcome. Also, as used herein, a
precursor may have a correlating relationship or a causal
relationship to the subsequent outcome.
[0022] As used herein, the term "sensors" and related terms, e.g.,
"sensors," refers to a device that is attached to a physical system
or a component of a physical system that may determine sensor
measurements, i.e., measurement data, physical system or the
component for a given point in time. Also, as used herein, sensors
facilitate the detection of sensor measurements and the
transmission of the sensor measurements to the computing
device.
[0023] As used herein, the term "sensor measurement" and related
terms, e.g., "sensor measurements," refers to a type of measurement
data that is sensed by a sensor or a plurality of sensors. The
sensor measurements may include, without limitation, data on the
mechanical integrity of a component, data on the mechanical
operation of a component, data on the chemical state of a
component, data on the electrical conductivity of a component, data
on the radiation signatures of a component, and data on the
temperature of a component. Sensor measurement data may also have
been detected previously and represent historical sensor
measurement data. Sensor measurement data may further have been
detected externally and imported into the system.
[0024] As used herein, the term "feature" and related terms, e.g.,
"feature library," refers to characteristics of sensor measurements
that are of interest in the analysis of the plurality of sensor
measurements. Also, as used herein, features facilitate finding
precursors to a failure for a particular type of component in the
physical system.
[0025] As used herein, the term "multivariate fusion" and related
terms, e.g., "multivariate fusion analysis," refers to the
observation and analysis of multiple variables at one time. Also,
as used herein, multivariate fusion involves bringing sensor
measurements from the plurality of sensor measurements into a
grouping and simultaneously analyzing all sensor measurements.
Additionally, as applied herein, multivariate fusion facilitates
determining features that are of interest, creating a predictive
model of the physical system, and designating at least one sensor
measurement used as an input to the predictive model as a precursor
to failure. Further, multivariate fusion may be used to observe and
analyze multiple features received from a single sensor. For
example, a single sensor may produce a plurality of sensor
measurements or a vector comprising sensor measurements. In this
case, observing and analyzing features from the single sensor can
incorporate multivariate fusion.
[0026] As used herein, the term "univariate analysis" and related
terms, e.g., "univariate diagnostic index" or "univariate
prognostic index," refers to the observation and analysis of a
single variable at one time, in contrast to multivariate fusion.
Also, as used herein, univariate fusion involves looking at sensor
measurements from a particular sensor and analyzing these sensor
measurements in relation to the physical system. Additionally, as
applied herein, univariate analysis facilitates creating a
predictive model of the physical system.
[0027] As used herein, the term "Bayesian analysis" and related
terms, e.g., "Bayesian inferences" and "naive Bayesian
classification," refer to a method of inference which considers the
probability of an event in light of a prior probability and a
likelihood function derived from existing relevant data. More
specifically, Bayesian analysis considers a set of data preceding
an outcome, determines what data from that set of data is relevant,
and determines an outcome probability based upon the general
likelihood of an outcome and the likelihood considering the
relevant set of data. Bayesian analysis allows for the constant
updating of a predictive model with new sets of evidence. Many
known models for applying Bayesian analysis exist including naive
Bayesian classification Bayesian log-likelihood functions. Also, as
used herein, Bayesian analysis facilitates distinguishing which
sensor measurements are most associated with the failure outcome
being evaluated, and distinguishing which sensors are therefore
most determinative to such an outcome.
[0028] Approximating language, as used herein throughout the
specification and claims, may be applied to modify any quantitative
representation that could permissibly vary without resulting in a
change in the basic function to which it is related. Accordingly, a
value modified by a term or terms, such as "about" and
"substantially", are not to be limited to the precise value
specified. In at least some instances, the approximating language
may correspond to the precision of an instrument for measuring the
value. Here and throughout the specification and claims, range
limitations may be combined and/or interchanged, such ranges are
identified and include all the sub-ranges contained therein unless
context or language indicates otherwise.
[0029] FIG. 1 is a simplified block diagram of a portion of an
exemplary computer-implemented system 100 for identifying a
precursor to a failure of a particular type of component in a
physical system. Computer-implemented system 100 includes a
physical system 105 composed of a plurality of components 107. In
the exemplary embodiment, physical system 105 is a locomotive and
the plurality of components 107 are components of locomotives
including, without limitation, locomotive engines, locomotive
wheels, locomotive electronics, locomotive brakes, locomotive
heating systems, locomotive cooling systems, and locomotive
communications systems. In alternative embodiments, physical system
105 can be any physical system 105 including plurality of
components 107 and capable of being monitored by a plurality of
sensors 110. These alternative embodiments of physical systems 105
may include, without limitation, vehicles, transportation systems,
manufacturing facilities, chemical processing facilities, power
generation facilities, infrastructure systems, and communication
systems. Physical system 105 is coupled to sensors 110. Also, in
the exemplary embodiment, sensors 110 are coupled to the wheels,
engine, and brakes of physical system 105 represented as a
locomotive. In alternative embodiments, sensors 110 can be coupled
to any component of the plurality of components 107 of physical
system 105.
[0030] Computer-implemented system 100 also includes a computing
device 130. Computing device 130 includes a processor 135 and a
memory device 140. Processor 135 and memory device 140 are coupled
to one another. Moreover, in the exemplary embodiment, computing
device 130 includes one processor 135 and one memory device 140. In
alternative embodiments, computing device 130 may include a
plurality of processors 135 and a plurality of memory devices 140.
Computing device 130 is associated with a database 150.
Furthermore, in the exemplary embodiment, database 150 is
manifested as a single database instance. In alternative
embodiments, database 150 is manifested as a plurality of database
instances.
[0031] Moreover, computing device 130 is configured to receive
sensor measurements 120 associated with physical system 105 from
sensors 110. In the exemplary embodiment, sensor measurements 120
include vibration data, rotational data, and thermal data from
plurality of components 107. In alternative embodiments, sensor
measurements 120 may include, without limitation, data on the
mechanical integrity of a component, and data on the mechanical
operation of a component. Also, in other alternative embodiments,
sensor measurements 120 may include data on the chemical state of a
component, data on the electrical conductivity of a component, data
on the radiation signatures of a component, and component thermal
data. Further, in additional alternative embodiments, sensor
measurements 120 may include a range of time which includes
multiple repair events.
[0032] In addition, computing device 130 is also configured to
store sensor measurements 120 at memory device 140. Computing
device 130 receives a plurality of sensor measurements 120 stored
at memory device 140. Computing device 130 is configured to receive
expert user input (not shown in FIG. 1) associated with an expert
user 155. Such input includes expert data representing information
obtained by human experts regarding the relationship between
sensors 110 and physical system 105.
[0033] Furthermore, computer-implemented system 100 includes a
monitoring system 160. As used herein, the term "monitoring system"
includes any programmable system including systems and
microcontrollers, reduced instruction set circuits, application
specific integrated circuits, programmable logic circuits, and any
other circuit capable of executing the monitoring functions
described herein. Monitoring systems may include sufficient
processing capabilities to execute support applications including,
without limitation, a Supervisory, Control and Data Acquisition
(SCADA) system and a Data Acquisition System (DAS). The above
examples are exemplary only, and thus are not intended to limit in
any way the definition and/or meaning of the term processor.
Monitoring system 160 is associated with, and capable of monitoring
and communicating with, physical system 105. Monitoring system 160
is also capable of communicating with computing device 130.
[0034] In operation, computing device 130 generates a predictive
model 170. Computing device 130 generates predictive model 170
using sensor measurements 120 and uses sensors 110 as inputs. In
alternative computer-implemented systems 100, computing device 130
generates predictive model 170 using sensor measurements 120 such
that a subset of sensors 110 in sensor measurements 120 is used as
inputs. In other computer-implemented systems 100, computing device
130 generates predictive model 170 using sensor measurements 120
which includes at least some expert user input received from expert
user 155 at computing system 130.
[0035] Also, in operation, computing device 130 designates at least
one designated sensor measurement 145 to be used as an input to
predictive model 170 as a precursor to failure of a particular type
of component 107 in physical system 105. Computing device 130
designates designated sensor measurement 145 and updates predictive
model 170 and stores designated sensor measurement 145 in memory
device 140 and/or database 150. In at least some
computer-implemented systems 100, computer device 130 designates
designated sensor measurement 145 and transmits designated sensor
measurement 145 and at least one mathematical operation of
predictive model 170 to monitoring system 160. In such
computer-implemented systems 100, monitoring system 160 monitors
physical system 105 for sensor measurements 120 that indicate
physical system 105 is approaching the end of its remaining useful
life.
[0036] FIG. 2 is a block diagram of computing device 130 used for
identifying a precursor to a failure of a particular type of
component 107 in physical system 105 (both shown in FIG. 1).
Computing device 130 includes a memory device 140 and a processor
135 operatively coupled to memory device 140 for executing
instructions. Processor 135 may include one or more processing
units. In some embodiments, executable instructions are stored in
memory device 140. Computing device 130 is configurable to perform
one or more operations described herein by programming processor
135. For example, processor 135 may be programmed by encoding an
operation as one or more executable instructions and providing the
executable instructions in memory device 140.
[0037] In the exemplary embodiment, memory device 140 is one or
more devices that enable storage and retrieval of information such
as executable instructions and/or other data. Memory device 140 may
include one or more tangible, non-transitory computer-readable
media, such as, without limitation, random access memory (RAM),
dynamic random access memory (DRAM), static random access memory
(SRAM), a solid state disk, a hard disk, read-only memory (ROM),
erasable programmable ROM (EPROM), electrically erasable
programmable ROM (EEPROM), and/or non-volatile RAM (NVRAM) memory.
The above memory types are exemplary only, and are thus not
limiting as to the types of memory usable for storage of a computer
program.
[0038] Memory device 140 may be configured to store sensor
measurements 120 (shown in FIG. 1) including, without limitation,
vibration data, chemical data, thermal data, electrical data,
and/or any other type of data. In some embodiments, processor 135
removes or "purges" data from memory device 140 based on the age of
the data. For example, processor 135 may overwrite previously
recorded and stored data associated with a subsequent time and/or
event. In addition, or alternatively, processor 135 may remove data
that exceeds a predetermined time interval. Also, memory device 140
includes, without limitation, sufficient data, algorithms, and
commands to facilitate identifying a precursor to a failure of a
particular type of component 107 in a physical system 105
(discussed below).
[0039] In some embodiments, computing device 130 includes a user
input interface 230. In the exemplary embodiment, user input
interface 230 is coupled to processor 135 and receives input from
expert user 155. User input interface 230 may include, for example,
a keyboard, a pointing device, a mouse, a stylus, a touch sensitive
panel, including, e.g., without limitation, a touch pad or a touch
screen, and/or an audio input interface, including, e.g., without
limitation, a microphone. A single component, such as a touch
screen, may function as both a display device of presentation
interface 220 and user input interface 230.
[0040] A communication interface 235 is coupled to processor 135
and is configured to be coupled in communication with one or more
other devices, such as a sensor or another computing device 130,
and to perform input and output operations with respect to such
devices. For example, communication interface 235 may include,
without limitation, a wired network adapter, a wireless network
adapter, a mobile telecommunications adapter, a serial
communication adapter, and/or a parallel communication adapter.
Communication interface 235 may receive data from and/or transmit
data to one or more remote devices. For example, a communication
interface 235 of one computing device 130 may transmit an alarm to
the communication interface 235 of another computing device 130.
Communications interface 235 facilitates machine-to-machine
communications, i.e., acts as a machine-to-machine interface.
[0041] Presentation interface 220 and/or communication interface
235 are both capable of providing information suitable for use with
the methods described herein, e.g., to expert user 155 or another
device. Accordingly, presentation interface 220 and communication
interface 235 may be referred to as output devices. Similarly, user
input interface 230 and communication interface 235 are capable of
receiving information suitable for use with the methods described
herein and may be referred to as input devices. In some
embodiments, expert user 155 uses presentation interface 220 and/or
communication interface 235 to input expert user input (not shown
in FIG. 2) into computing system 130. In at least some other
embodiments user expert 155 uses presentation interface 220 and/or
communication interface 235 to review a plurality of candidate
models determining precursors to failure for particular type of
component 107 in physical system 105.
[0042] FIG. 3 is a flow chart of an exemplary process 300 of the
flow of information in computer-based system 100 (shown in FIG. 1).
In the exemplary embodiment, expert user input 310 associated with
expert user 155 (shown in FIG. 1) and historical data 305 are
received from memory device 140 (shown in FIG. 1). Historical data
305 is representative of historical sensor measurements received as
sensor measurements 120 (shown in FIG. 1). Feature extraction 315
is performed upon expert user input 310 associated with expert user
155 (shown in FIG. 1) and upon historical data 305. Feature
extraction 315 represents selecting data from expert user input 310
and historical data 305 and preparing selected data for processing.
In at least some embodiments, a pre-defined feature library 320 is
applied to the extracted feature data 315. In the at least some
embodiments, pre-defined feature library 320 is used to
pre-determine which features are likely to be more or less relevant
to predictive model 170 (shown in FIG. 1).
[0043] Additionally, features are selected 325 from features
extracted 315. In the exemplary embodiment, feature selection 325
substantially represents generating predictive model 170 to
determine a precursor to failure of a particular component in
physical system 105 (shown in FIG. 1). Feature selection may
include, without limitation, Bayesian analysis, log-likelihood
analysis, adaptive modeling, and any other mathematical or
computational operation capable of determining which feature 325
may be a precursor to a failure of a particular component in
physical system 105 (shown in FIG. 1).
[0044] Furthermore, the method determines 330 whether features 325
selected are sufficiently distinct to identify precursors. In the
exemplary embodiment, distinction can be set by, without
limitation, a threshold within the system, a standard system
requirement of prediction quality, or expert user 155 determined
requirement (not shown) of prediction quality. If features selected
325 are not determined 330 to be sufficiently distinct, the process
is repeated 340. If features 325 selected are determined 330 to be
sufficiently distinct, feature 325 is identified as a precursor 335
and can be associated to sensor measurements 120 detected by
designated sensor 145 (shown in FIG. 1).
[0045] FIG. 4 is a flow chart of an exemplary method 400 for
identifying a precursor to a failure of a particular type of
component 107 in physical system 105 (both shown in FIG. 1) using
process 300 (shown in FIG. 3). Historical data 305 (shown in FIG.
3) is received 415 from memory device 140 (shown in FIG. 1). In the
exemplary embodiment, historical data 305 includes data received
from sensors 110 (shown in FIG. 1) as sensor measurements 120
(shown in FIG. 1) obtained from physical system 105. In alternative
embodiments, historical data 305 further includes sensor
measurements 120 obtained from physical systems distinct from, but
similar to, physical system 105. In other embodiments, historical
data 305 further includes sensor measurements 120 obtained from
simulations of physical system 105.
[0046] Also, predictive model 170 (shown in FIG. 1) is generated
420 using sensor measurements 120 in historical data 305 as inputs.
In at least some embodiments, generating 420 predictive model 170
uses a subset of sensor measurements 120 in historical data 305 as
inputs. In other embodiments, generating 420 predictive model 170
uses a single sensor measurement 120 as an input and conducts a
univariate analysis. The univariate analysis may include, without
limitation, any mathematical function of single sensor measurement
120.
[0047] Further, in the exemplary embodiment, generating 420
predictive model 170 involves combining at least two sensor
measurements 120 in a mathematical operation. Mathematical
operation generally involves a process of multivariate fusion where
multiple sensor measurements 120 are evaluated as outcome variables
simultaneously. Multivariate fusion may involve, without
limitation, factor analysis, polynomial equations, adaptive
modeling, or any other known or discovered mathematical
operation.
[0048] Moreover, in at least some embodiments, historical data 305
received spans multiple repair events. In these embodiments,
generating 420 predictive model 170 involves generating a plurality
of candidate predictive models (not shown) where the plurality of
candidate predictive models use a random selection of sensor
measurements 120 as inputs. Also, in the at least some embodiments,
generating 420 predictive model 170 further involves determining
which of the plurality of candidate predictive models most
accurately differentiates between sensor measurements 120 taken
before a repair event and sensor measurements 120 taken after a
repair event. Further, these embodiments, generating 420 predictive
model 170 also involves designating as predictive model 170 the
most accurate of the plurality of candidate predictive models.
[0049] Furthermore, in some embodiments, historical data 305
includes expert user input 310 (shown in FIG. 3) associated with
expert user 155 (shown in FIG. 1). In these embodiments, generating
420 predictive model 170 also involves using at least some expert
user input 310 associated with expert user 155. Such expert user
input 310 associated with expert user 155 may include, without
limitation, specific sensor measurements 120 designated as related
to each other by expert user 155, specific sensor measurements 120
designated as unrelated by expert user 155, and combinations of
specific sensor measurements 120 designated as related to each
other by expert user 155. In these embodiments, generating 420
predictive model 170 involves distinguishing the significance of
expert user input 310 associated with expert user 155 from other
historical data 305. Distinguishing may be accomplished by methods
including, without limitation, multivariate fusion, Bayesian
analysis, and the use of feature libraries.
[0050] Also, in the exemplary embodiment, generating 420 a
predictive model 170 involves feature selection 325 (shown in FIG.
3) to distinguish which sensor measurements 120 are precursors to
the failure of the particular type of component 107. In at least
some embodiments, feature selection 325 includes the use of
pre-defined feature library 320 (shown in FIG. 3) which is stored
in database 150 (shown in FIG. 1). Pre-defined feature library 320
facilitates the identification and selection of features which
facilitates generating 420 predictive model 170.
[0051] Further, at least one designated sensor measurement 145
(shown in FIG. 1) used as an input to a predictive model 170 is
designated 425 as a precursor to failure. In at least some
embodiments, a combination of designated sensor measurements 145
are designated as a precursor to failure. In alternative
embodiments, designating 425 at least one designated sensor
measurement 145 involves transmitting the designation of at least
one designated sensor measurement 145 to monitoring system 160
(shown in FIG. 1) which may monitor physical system 105.
[0052] The computer-implemented systems and methods as described
herein facilitate increasing the remaining useful life of a
physical system. Also, such systems and methods facilitate reducing
the cost of servicing the physical system. Further, such systems
and methods facilitate improving the monitoring of the physical
system by identifying sensors that are precursors to failure for a
particular type of component in the physical system.
[0053] A technical effect of systems and methods described herein
includes at least one of: (a) enhancing the remaining useful life
of physical systems by enabling monitoring of the most important
sensors for failure of components in the physical system; (b)
reducing the time to identify a failure of components in the
physical system by focusing on the most important sensors for
failure of components in the physical system; and (c) facilitating
identification of precursors to failure by expediting the analysis
of complex sensor data as precursors to failure for components of
the physical system.
[0054] Exemplary embodiments of computer-implemented systems and
methods for identifying a precursor to a failure of a component in
a physical system are described above in detail. The
computer-implemented systems and methods of operating such systems
are not limited to the specific embodiments described herein, but
rather, components of systems and/or steps of the methods may be
utilized independently and separately from other components and/or
steps described herein. For example, the methods may also be used
in combination with other enterprise systems and methods, and are
not limited to practice with only the methods and systems for
identifying a precursor to a failure of a component in a physical
system, as described herein. Rather, the exemplary embodiment can
be implemented and utilized in connection with many other
enterprise applications.
[0055] Although specific features of various embodiments of the
invention may be shown in some drawings and not in others, this is
for convenience only. In accordance with the principles of the
invention, any feature of a drawing may be referenced and/or
claimed in combination with any feature of any other drawing.
[0056] This written description uses examples to disclose the
invention, including the best mode, and also to enable any person
skilled in the art to practice the invention, including making and
using any devices or systems and performing any incorporated
methods. The patentable scope of the invention is defined by the
claims, and may include other examples that occur to those skilled
in the art. Such other examples are intended to be within the scope
of the claims if they have structural elements that do not differ
from the literal language of the claims, or if they include
equivalent structural elements with insubstantial differences from
the literal language of the claims.
* * * * *